Data compensation method of radio frequency chip

Abstract

The invention relates to the technical field of radio frequency chip data compensation, and discloses a radio frequency chip data compensation method, which comprises the following steps: acquiring I / Q signal state data of a radio frequency chip in real time; calculating a first time delay compensation parameter based on a fuzzy control algorithm; constructing a nonlinear compensation filter to obtain a preliminary time delay compensation signal and performing feedback adjustment on the first time delay compensation parameter; constructing a long short-term memory network to predict and obtain a third time delay compensation parameter; and inputting the third time delay compensation parameter into a nonlinear compensation filter. In the prior art, a fixed interpolation or simple linear feedback method is adopted, and especially under the condition that large-scale MIMO and multi-channel ADC clocks are inconsistent and the signal state is dynamically changed, high-precision, real-time and dynamic time delay compensation is difficult to realize. According to the invention, by introducing the fuzzy control algorithm and combining the long and short term network, accurate prediction and real-time updating of the time delay compensation parameters are realized, and the data synchronization performance of the radio frequency chip is improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of radio frequency chip data compensation, and particularly relates to a data compensation method for radio frequency chips. Background Art

[0002] Currently, with the rapid development of 5G, 6G and future communication systems, radio frequency chips are increasingly widely used in fields such as wireless communication, radar systems, and the Internet of Things. Especially in large-scale MIMO (Massive Multiple-Input Multiple-Output) systems, to meet the requirements of high throughput, low latency, and high reliability, radio frequency chips have extremely high requirements for signal synchronization and data accuracy during multi-channel parallel processing. However, due to the heterogeneity of radio frequency signal channels and the inconsistency of ADC (analog-to-digital converter) clocks for each channel, there are generally subtle but significant fractional delay differences in the system. Although this delay difference is small, in an environment of high-density and multi-channel parallel transmission, it will cause the synchronization of I / Q signals to be out of adjustment, which will in turn lead to signal demodulation errors, increased phase noise, and a decrease in data processing accuracy. For example, in traditional large-scale MIMO systems, methods such as fixed interpolation or piecewise polynomials are often used to achieve delay compensation, but these methods have deficiencies, including: poor real-time performance, fixed or segmented methods usually rely on pre-set compensation parameters and cannot respond in real time to the dynamic changes of the channel state, resulting in large synchronization errors. High computational complexity, when facing a large number of parallel channels, algorithms such as piecewise polynomial interpolation have high computational complexity and are difficult to achieve fast calculation under the requirements of low power consumption and high real-time performance. Insufficient adaptability, traditional methods mostly use linear models or fixed rules and are difficult to fully capture the non-linear distortion and multi-dimensional information in the signal, resulting in poor compensation effects and serious error accumulation problems in dynamic and complex channel environments. Therefore, there is an urgent need for a delay compensation method that can still achieve real-time and accurate delay compensation in complex and dynamic channel environments to improve the signal synchronization performance and data processing accuracy of radio frequency chips. Summary of the Invention

[0003] Aiming at the above-mentioned existing technical deficiencies, the purpose of the present invention is to propose a data compensation method for radio frequency chips, aiming to solve the technical problem that it is difficult to achieve high-precision, real-time dynamic delay compensation by using fixed interpolation or simple linear feedback methods in the prior art, especially under the conditions of large-scale MIMO and inconsistent ADC clocks for multi-channels and dynamic changes in signal states.

[0004] To solve the above technical problems, the present invention adopts the following technical solutions: The present invention provides a data compensation method for radio frequency chips,

[0005] The data compensation method for the radio frequency chip includes:

[0006] Step S10: Real-time collect the real-time status data of the I / Q signals of the RF chip. The real-time status data includes real-time signal amplitude, real-time signal frequency, and real-time noise level. Calculate the first time-delay compensation parameter based on the fuzzy control algorithm combined with the real-time status data.

[0007] Step S20: Construct a non-linear compensation filter. Input the first time-delay compensation parameter into the non-linear compensation filter to obtain a preliminary time-delay compensation signal.

[0008] Step S30: Calculate the signal synchronization error based on the first time-delay compensation signal. Feedback-adjust the first time-delay compensation parameter according to the signal synchronization error to obtain the second time-delay compensation parameter.

[0009] Step S40: Construct a long short-term memory network and perform pre-training. Dynamically predict the second time-delay compensation parameter through the pre-trained long short-term memory network to obtain the third time-delay compensation parameter.

[0010] Step S50: Apply the third time-delay compensation parameter. Input the third time-delay compensation parameter into the non-linear compensation filter to obtain the final time-delay compensation signal.

[0011] Preferably, in step S10, the step of real-time collecting the real-time status data of the I / Q signals of the RF chip, where the real-time status data includes real-time signal amplitude, real-time signal frequency, and real-time noise level, and calculating the first time-delay compensation parameter based on the fuzzy control algorithm combined with the real-time status data specifically includes:

[0012] Step S101: Real-time collect the real-time status data of the I / Q signals of the RF chip at time t from the RF chip, including real-time signal amplitude A(t), real-time signal frequency f(t), and real-time noise level N(t).

[0013] Step S102: Define three-layer membership degrees for the signal amplitude A(t), real-time signal frequency f(t), and real-time noise level N(t) respectively, and use the triangular method to quantify the signal amplitude A(t), real-time signal frequency f(t), and real-time noise level N(t) to form fuzzy variable inputs.

[0014] Step S103: Deduce according to the preset fuzzy control rule base using the preset Mamdani model to obtain a fuzzy set output.

[0015] Step S104: Perform defuzzification processing on the fuzzy set output, using the centroid method or the maximum membership degree method, and calculate the first time-delay compensation parameter.

[0016] Preferably, in step S20, the step of constructing a non-linear compensation filter, inputting the first time-delay compensation parameter into the non-linear compensation filter to obtain a preliminary time-delay compensation signal specifically includes:

[0017] Step S201: Design a non - linear filter based on the Volterra series expansion method;

[0018] Step S202: Input the first time - delay compensation parameter into the non - linear compensation filter, and perform weighted processing on the linear term and high - order non - linear term in the non - linear compensation filter to obtain the mapped first time - delay compensation parameter;

[0019] Step S203: Perform time - domain compensation on the mapped first time - delay compensation parameter to obtain a preliminary time - delay compensation signal.

[0020] Preferably, in step S20, the step of performing time - domain compensation on the mapped first time - delay compensation parameter to obtain a preliminary time - delay compensation signal adopts the formula: ;

[0021] where, is the preliminary time - delay compensation signal; is the mapped first time - delay compensation parameter; is a non - linear adjustment factor for further correcting the compensation effect according to the signal estimation error; is the input signal; is the estimated value of the input signal; is the integration variable; is the error integral term; is the delay correction term.

[0022] Preferably, in step S30, the step of calculating the signal synchronization error according to the first time - delay compensation signal and performing feedback adjustment on the first time - delay compensation parameter according to the signal synchronization error to obtain the second time - delay compensation parameter specifically includes:

[0023] Step S301: Obtain the target synchronization signal from the preset prior information, and perform the same - time - reference standardization processing on the target synchronization signal and the first time - delay compensation signal;

[0024] Step S302: Calculate the signal synchronization error by using the mean - square error according to the target synchronization signal and the first time - delay compensation signal after the standardization processing;

[0025] Step S303: Judge the direction and amplitude of the adjustment required for the time - delay compensation parameter according to the signal synchronization error in combination with the PID control method, and output a correction amount;

[0026] Step S304: Apply the correction amount to perform feedback adjustment on the first time - delay compensation parameter to obtain the second time - delay compensation parameter.

[0027] Preferably, in step S40, the steps of constructing a long short-term memory network and performing pre-training, and dynamically predicting the second time delay compensation parameter through the pre-trained long short-term memory network to obtain the third time delay compensation parameter specifically include:

[0028] Step S401: Obtain the historical time series data of the second time delay compensation parameter, the historical time series data of the signal amplitude, the historical time series data of the signal frequency, and the historical time series data of the noise level. Denoise, normalize, and time-align the collected time series data, and construct a fused feature vector.

[0029] Step S402: Construct a long short-term memory network, including an input layer, multiple LSTM layers, a fully connected layer, and an output layer. Among them, the input layer is used to receive the fused feature vector, and the multiple LSTM layers are used to capture the long-term dependencies and non-linear relationships in the time series data; the fully connected layer is used to map the output of the LSTM layer to the predicted value, and the output layer is used to output the predicted value.

[0030] Step S403: Obtain the target synchronization signal from the preset prior information, construct the standard compensation parameter output according to the target synchronization signal, and train the long short-term memory network using the fused feature vector and the standard compensation parameter output.

[0031] Step S404: Use the second time delay compensation parameter as the output of the long short-term memory network, and the long short-term memory network outputs the third time delay compensation parameter.

[0032] Preferably, in step S40, during the process of training the long short-term memory network, a synchronization performance index is also introduced as a regularization term to design the loss function L. The loss function L adopts the formula: ;

[0033] where i represents the i-th training sample, is the third time delay compensation parameter predicted by the long short-term memory network; is the standard compensation parameter output constructed according to the target synchronization signal; is the preset regularization coefficient, used to balance the two parts of the loss; is a measure of the signal synchronization error after compensation based on the predicted parameter; N is the preset number of training samples.

[0034] The present invention also provides a data compensation system for a radio frequency chip, including:

[0035] A real-time status data acquisition module, used to collect the real-time status data of the I / Q signal of the radio frequency chip in real time. The real-time status data includes the real-time signal amplitude, the real-time signal frequency, and the real-time noise level, and calculates the first time delay compensation parameter based on the fuzzy control algorithm combined with the real-time status data.

[0036] A fuzzy control time-delay parameter calculation module, which is used to construct a non-linear compensation filter, input the first time-delay compensation parameter into the non-linear compensation filter, and obtain a preliminary time-delay compensation signal;

[0037] A non-linear compensation filtering module, which is used to calculate the signal synchronization error according to the first time-delay compensation signal, and perform feedback adjustment on the first time-delay compensation parameter according to the signal synchronization error to obtain a second time-delay compensation parameter;

[0038] A synchronization error feedback and parameter adjustment module, which is used to construct a long short-term memory network and perform pre-training, and dynamically predict the second time-delay compensation parameter through the pre-trained long short-term memory network to obtain a third time-delay compensation parameter;

[0039] An LSTM time-delay prediction and optimization module, which is used to apply the third time-delay compensation parameter, input the third time-delay compensation parameter into the non-linear compensation filter, and obtain a final time-delay compensation signal.

[0040] The present invention also provides a computer program product, including a data compensation program for a radio frequency chip, and when the data compensation program for the radio frequency chip is executed by a processor, it implements the data compensation method for the radio frequency chip described above.

[0041] The beneficial effects of the present invention are as follows: compared with the prior art that uses fixed interpolation or simple linear feedback methods, especially under the conditions of inconsistent clocks of large-scale MIMO and multi-channel ADCs and dynamic changes in signal states, it is difficult to achieve high-precision and real-time dynamic time-delay compensation. By introducing a fuzzy control algorithm and combining it with a long short-term network, the present application realizes accurate prediction and real-time update of time-delay compensation parameters, and improves the data synchronization performance of the radio frequency chip. Description of the Drawings

[0042] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0043] Figure 1 It is a schematic flowchart of the first embodiment of a data compensation method for a radio frequency chip of the present invention.

[0044] Figure 2 It is a schematic diagram of the device for a data compensation method for a radio frequency chip of the present invention. Detailed Embodiments

[0045] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0046] Embodiment 1: As Figure 1 shown, it is a schematic flowchart of the first embodiment of the data compensation method for the radio frequency chip of the present invention, and the first embodiment of the data compensation method for the radio frequency chip of the present invention is proposed.

[0047] In the first embodiment, the data compensation method for the radio frequency chip includes:

[0048] Step S10: Real-time collect the real-time state data of the I / Q signal of the radio frequency chip. The real-time state data includes the real-time signal amplitude, real-time signal frequency, and real-time noise level. Calculate the first delay compensation parameter based on the fuzzy control algorithm in combination with the real-time state data;

[0049] It should be noted that in step S10, the step of real-time collecting the real-time state data of the I / Q signal of the radio frequency chip, where the real-time state data includes the real-time signal amplitude, real-time signal frequency, and real-time noise level, and calculating the first delay compensation parameter based on the fuzzy control algorithm in combination with the real-time state data specifically includes:

[0050] Step S101: Real-time collect the real-time state data of the I / Q signal of the radio frequency chip at time t from the radio frequency chip, including the real-time signal amplitude A(t), real-time signal frequency f(t), and real-time noise level N(t);

[0051] Step S102: Define three layers of membership degrees for the signal amplitude A(t), real-time signal frequency f(t), and real-time noise level N(t) respectively, and use the triangular method to quantify the signal amplitude A(t), real-time signal frequency f(t), and real-time noise level N(t) to form fuzzy variable inputs;

[0052] Step S103: Deduce according to the preset fuzzy control rule base using the preset Mamdani model to obtain the fuzzy set output;

[0053] Step S104: Perform defuzzification processing on the fuzzy set output, using the centroid method or the maximum membership degree method, and calculate the first delay compensation parameter.

[0054] It can be understood that by adopting the triangular method for quantization, the continuously varying signal state data can be mapped into discrete fuzzy sets, providing accurate and stable input data for subsequent fuzzy inference. Multiple rules are stored in the fuzzy control rule base, such as the rule "If the signal amplitude is high, the frequency is on the low side, and the noise is low, then the time-delay compensation parameter should be large", etc. These rules are comprehensively reflected in the output fuzzy set through the Mamdani model, ensuring that the algorithm can flexibly handle the non-linear changes of the signal. The defuzzification process concretizes the result of the fuzzy inference, making the output parameter have a clear numerical representation, which is convenient for the parameter input and time-delay compensation processing of subsequent modules (such as non-linear compensation filters).

[0055] It should be understood that based on the dynamic acquisition of signal state data, the present invention uses fuzzy control technology to comprehensively consider multi-dimensional factors such as signal amplitude, frequency, and noise, and calculates the first time-delay compensation parameter in real time. Compared with the traditional fixed or piecewise interpolation methods, this scheme has higher real-time performance, self-adaptability, and compensation accuracy, and can effectively cope with the signal synchronization problems caused by large-scale MIMO and ADC time-delay inconsistencies, thereby improving the demodulation accuracy and data processing accuracy of the entire communication.

[0056] For example, in a certain actual test, the signal data collected through step S101 is as follows: 、 GHz and ; After quantization in step S102, the corresponding membership degrees are respectively determined as "medium", "relatively low", and "low". In step S103, based on the preset fuzzy rules, the fuzzy output is "the time-delay compensation parameter is large". Finally, through the centroid method for defuzzification, the specific value of the first time-delay compensation parameter is calculated ns. This result shows that the present method can accurately estimate the required time-delay compensation parameter in a dynamic channel environment, thus providing an accurate basis for subsequent compensation.

[0057] Step S20: Construct a non-linear compensation filter, input the first time-delay compensation parameter into the non-linear compensation filter, and obtain a preliminary time-delay compensation signal;

[0058] It should be noted that in step S20, the steps of constructing a non-linear compensation filter and inputting the first time-delay compensation parameter into the non-linear compensation filter to obtain a preliminary time-delay compensation signal specifically include:

[0059] Step S201: Design a non-linear filter based on the Volterra series expansion method;

[0060] Step S202: Input the first time-delay compensation parameter into the non-linear compensation filter, and perform weighted processing on the linear term and the high-order non-linear term in the non-linear compensation filter to obtain the mapped first time-delay compensation parameter;

[0061] Step S203: Perform time-domain compensation on the mapped first time-delay compensation parameter to obtain a preliminary time-delay compensation signal.

[0062] The step of performing time-domain compensation on the mapped first time-delay compensation parameter to obtain a preliminary time-delay compensation signal adopts the formula: ;

[0063] where is the preliminary time-delay compensation signal; is the mapped first time-delay compensation parameter; is the non-linear adjustment factor, which is used to further correct the compensation effect according to the signal estimation error; is the input signal; is the estimated value of the input signal; is the integration variable; is the error integral term; is the delay correction term.

[0064] It can be understood that the above steps introduce the Volterra series expansion method into the design of the non-linear filter, enabling the full description and compensation of the small non-linear time-delay differences between ADCs; at the same time, by performing weighted processing on the linear term and the high-order non-linear term respectively, the input first time-delay compensation parameter is flexibly mapped, providing a more accurate correction basis for time-domain compensation. The time-domain compensation processing formula not only directly corrects the delay of the input signal, but also introduces a non-linear adjustment factor and an error integral term, comprehensively considering the historical error accumulation, thereby finely correcting the delay to ensure that the finally obtained preliminary compensation signal is closer to the desired synchronization signal.

[0065] It should be understood that the non-linear filter designed by the present invention using the Volterra series expansion obtains a more accurate preliminary time-delay compensation signal after weighted mapping, time-domain compensation and online optimization of the first time-delay compensation parameter. Compared with the traditional method, this solution has higher compensation accuracy, better adaptability and real-time performance in dealing with the non-linear problems caused by the inconsistent ADC time delays in large-scale MIMO systems, thus significantly improving the signal synchronization performance and data processing quality of the radio frequency chip.

[0066] For example, in a test, assume that the first time-delay compensation parameter obtained from step S10 ns, after being weighted and processed in step S202, is mapped to ns. At the same time, the measured delay correction term ns, input signal and its estimated value The integral term of the error between them is calculated through integration to obtain a certain value. After being corrected by the non - linear adjustment factor finally, the preliminary compensation signal obtained in step S203 significantly reduces the synchronization error. Its error index is reduced by more than about 20% compared with the traditional method using fixed compensation parameters, verifying the efficient compensation ability of this scheme for complex non - linear time - delay errors in a real - time dynamic environment.

[0067] Step S30: Calculate the signal synchronization error according to the first time - delay compensation signal, and perform feedback adjustment on the first time - delay compensation parameter according to the signal synchronization error to obtain the second time - delay compensation parameter;

[0068] It should be noted that in step S30, the steps of calculating the signal synchronization error according to the first time - delay compensation signal and performing feedback adjustment on the first time - delay compensation parameter according to the signal synchronization error to obtain the second time - delay compensation parameter specifically include:

[0069] Step S301: Obtain the target synchronization signal from the preset prior information, and perform the same - time - reference - based processing on the target synchronization signal and the first time - delay compensation signal;

[0070] Step S302: Calculate the signal synchronization error by using the mean - square error according to the target synchronization signal and the first time - delay compensation signal after the reference - based processing;

[0071] Step S303: Determine the direction and amplitude of the adjustment required for the time - delay compensation parameter according to the signal synchronization error in combination with the PID control method, and output a correction amount;

[0072] Step S304: Apply the correction amount to perform feedback adjustment on the first time - delay compensation parameter to obtain the second time - delay compensation parameter.

[0073] It can be understood that in a large - scale MIMO system, due to the inconsistency of ADC clocks and the dynamic change of the channel environment, the first time - delay compensation signal often cannot fully achieve the desired synchronization effect. Therefore, it is necessary to calculate the actual signal synchronization error and adjust the first time - delay compensation parameter based on the error by using a feedback control method, so as to obtain a more accurate second time - delay compensation parameter and provide an accurate input for subsequent compensation.

[0074] It should be understood that the mean square error metric can comprehensively reflect the synchronization deviation of the signal over a period of time, thereby obtaining a more stable and accurate error index. The PID control method has the advantages of rapid adjustment and smooth response, and can calculate the appropriate correction amount in real time, making the compensation parameters quickly tend to the optimal state. Applying the correction amount to feedback-adjust the first time-delay compensation parameter to obtain the second time-delay compensation parameter. The feedback adjustment mechanism enables dynamic correction of the compensation deviation caused by environmental changes or imperfect models, so that the time-delay compensation parameter continuously approaches the optimal value. In summary, using real-time feedback and the PID control method to correct the first time-delay compensation parameter can not only make up for the deficiencies of traditional fixed compensation methods, but also continuously optimize the compensation effect in a dynamic and complex communication environment, ensuring the high precision and stability of radio frequency chip data synchronization.

[0075] For example, in a certain actual test, the target synchronization signal is obtained through step S301 and the first time-delay compensation signal After being processed by time benchmarking and compared, it is found that within the past 10 milliseconds, the mean square error between the two is 0.05. Using the pre-set PID parameters 、 and After that, the correction amount ns is calculated. Adding this correction amount to the first time-delay compensation parameter ns later, the second time-delay compensation parameter ns is obtained. The experimental results show that after feedback adjustment, the signal synchronization error is reduced by about 15%, significantly improving the signal synchronization effect.

[0076] Step S40: Construct a long short-term memory network and perform pre-training, and dynamically predict the second time-delay compensation parameter through the pre-trained long short-term memory network to obtain the third time-delay compensation parameter;

[0077] It should be noted that in step S40, the steps of constructing a long short-term memory network and performing pre-training, and dynamically predicting the second time-delay compensation parameter through the pre-trained long short-term memory network to obtain the third time-delay compensation parameter specifically include:

[0078] Step S401: Obtain the historical second time-delay compensation parameter time series data, historical signal amplitude time series data, historical signal frequency time series data and historical noise level time series data, perform denoising, normalization and time alignment processing on the collected time series data, and construct a fused feature vector;

[0079] Step S402: Construct a long short-term memory network, including an input layer, multiple LSTM layers, a fully connected layer, and an output layer; among them, the input layer is used to receive the fused feature vector, and the multiple LSTM layers are used to capture the long-term dependencies and non-linear relationships in the time-series data; the fully connected layer is used to map the output of the LSTM layer to the predicted value, and the output layer is used to output the predicted value;

[0080] Step S403: Obtain the target synchronization signal from the preset prior information, construct the standard compensation parameter output according to the target synchronization signal, and use the fused feature vector and the standard compensation parameter output to train the long short-term memory network;

[0081] Step S404: Use the second time-delay compensation parameter as the output of the long short-term memory network, and the long short-term memory network outputs the third time-delay compensation parameter.

[0082] Among them, during the process of training the long short-term memory network, it also includes introducing the synchronization performance index as a regularization term to design the loss function L, and the loss function L adopts the formula: ;

[0083] where i represents the i-th training sample, is the third time-delay compensation parameter predicted by the long short-term memory network; is the standard compensation parameter output constructed according to the target synchronization signal; is the preset regularization coefficient, which is used to balance the two parts of the loss; is the measure of the signal synchronization error after compensation based on the prediction parameter; N is the preset number of training samples.

[0084] It can be understood that traditional methods often only use single-parameter prediction, while this step can more comprehensively reflect the channel dynamics and compensation requirements by fusing multi-dimensional features, thus providing rich information for subsequent prediction. Using a deep LSTM network can fully learn the time-series dependencies in historical data, far superior to traditional simple linear or polynomial models, and can more accurately capture non-linear dynamic changes. By introducing the synchronization performance index as a regularization term, the model training not only focuses on the numerical difference between the predicted value and the true value, but also directly optimizes the signal synchronization performance, which can significantly reduce the compensation error in practice and is significantly better than the traditional method that only considers MSE. The online inference of LSTM can dynamically predict the time-delay compensation parameter in real time by combining the latest data, with a fast response speed and high accuracy, and can better adapt to environmental changes compared with the prediction of traditional fixed models. The adaptive update mechanism ensures that the LSTM model can continuously adapt to the changing channel conditions, avoids the decline of the model's prediction ability, ensures long-term high-precision time-delay compensation, and has higher robustness compared with the traditional once-trained model.

[0085] It should be understood that in this embodiment, the LSTM network is used to perform deep learning and prediction on historical time-series data, and the third delay compensation parameter is dynamically output, thereby achieving precise control of delay compensation in a complex and changeable communication environment. This solution is superior to traditional methods in terms of accuracy, real-time performance, and self-adaptability, providing higher technical guarantee for data compensation of radio frequency chips.

[0086] Step S50: Apply the third delay compensation parameter, input the third delay compensation parameter into the non-linear compensation filter, and obtain the final delay compensation signal.

[0087] It should be understood that by using the third delay compensation parameter for compensation, the cumulative error caused by noise, channel fluctuation, and ADC difference can be overcome, and dynamic and adaptive delay correction can be achieved. Compared with the fixed compensation parameter of the traditional method, this method has higher real-time performance and adaptability, thus significantly improving the signal synchronization accuracy and the overall data processing quality.

[0088] Embodiment 2: In addition, a data compensation system for a radio frequency chip provided by the present invention adopts a data compensation method for a radio frequency chip in the above embodiment, and can solve the technical problem of data compensation for a radio frequency chip. Compared with the prior art, the beneficial effects of the data compensation system for a radio frequency chip provided by the present invention are the same as those of the data compensation method for a radio frequency chip provided in the above embodiment, and other technical features in the data compensation system for a radio frequency chip are the same as those disclosed in the above embodiment method, and will not be elaborated here.

[0089] Embodiment 3: The present invention provides a data compensation device for a radio frequency chip. Please refer to Figure 2, A data compensation device for a radio frequency chip includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to execute a data compensation method for a radio frequency chip in the first embodiment above. The data compensation device for a radio frequency chip in the embodiments of the present invention may include, but is not limited to, mobile terminals such as mobile phones, laptop computers, digital broadcast receivers, PDAs (Personal Digital Assistant), PADs (Portable Application Description: tablet computers), PMPs (Portable Media Player: portable multimedia players), vehicle terminals (such as vehicle navigation terminals), etc., and fixed terminals such as digital TVs, desktop computers, etc. A data compensation device for a radio frequency chip is merely an example and should not impose any limitations on the functions and scope of use of the embodiments of the present invention. A data compensation device for a radio frequency chip may include a processing device 1001 (such as a central processing unit, a graphics processing unit, etc.), which may perform various appropriate actions and processes according to a program stored in a read-only memory (ROM: Read Only Memory) 1002 or a program loaded from a storage device 1003 into a random access memory (RAM: Random Access Memory) 1004. In the RAM 1004, various programs and data required for the operation of a data compensation device for a radio frequency chip are also stored. The processing device 1001, the ROM 1002, and the RAM 1004 are connected to each other via a bus 1005. An input / output (I / O) interface 1006 is also connected to the bus. Generally, the following systems may be connected to the I / O interface 1006: an input device 1007 including, for example, a touch screen, a touchpad, a keyboard, a mouse, an image sensor, a microphone, an accelerometer, a gyroscope, etc.; an output device 1008 including, for example, a liquid crystal display (LCD: Liquid Crystal Display), a speaker, a vibrator, etc.; a storage device 1003 including, for example, a magnetic tape, a hard disk, etc.; and a communication device 1009. The communication device 1009 may allow a data compensation device for a radio frequency chip to communicate with other devices wirelessly or wireline to exchange data. Although the figure shows a data compensation device for a radio frequency chip having various systems, it should be understood that it is not required to implement or have all the systems shown. Instead, more or fewer systems may be implemented or had.

[0090] Embodiment 4: The present invention further provides a computer program product, including a computer program, and when the computer program is executed by a processor, it implements the steps of a data compensation method for a radio frequency chip as described above. The computer program product provided by the present invention can solve the technical problem of data compensation for a radio frequency chip. Compared with the prior art, the beneficial effects of the computer program product provided by the present invention are the same as those of the data compensation method for a radio frequency chip provided in the above embodiment, and will not be elaborated here.

[0091] Specifically, according to the embodiments disclosed by the present invention, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, an embodiment disclosed by the present invention includes a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program contains program codes for executing the methods shown in the flowcharts. In such an embodiment, the computer program can be downloaded and installed from the network through a communication device, or installed from the storage device 1003, or installed from the ROM 1002. When the computer program is executed by the processing device 1001, it executes the above functions defined in the methods of the embodiments disclosed by the present invention.

[0092] It should be understood that the various parts disclosed by the present invention can be implemented by hardware, software, firmware, or a combination thereof. In the description of the above embodiments, specific features, structures, materials, or characteristics can be combined in a suitable manner in any one or more embodiments or examples.

[0093] Obviously, those skilled in the art can make various modifications and variations to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalent technologies, the present invention also intends to include these modifications and variations.

Claims

1. A data compensation method for a radio frequency chip, characterized in that The method includes: Step S10: Real-time collect the real-time status data of the I / Q signals of the RF chip. The real-time status data includes real-time signal amplitude, real-time signal frequency, and real-time noise level. Calculate the first time-delay compensation parameter based on the fuzzy control algorithm combined with the real-time status data; Step S20: Construct a non-linear compensation filter, input the first time-delay compensation parameter into the non-linear compensation filter to obtain a preliminary time-delay compensation signal; Step S30: Calculate the signal synchronization error according to the first time-delay compensation signal, and perform feedback adjustment on the first time-delay compensation parameter according to the signal synchronization error to obtain the second time-delay compensation parameter; Step S40: Construct a long short-term memory network and perform pre-training. Dynamically predict the second time-delay compensation parameter through the pre-trained long short-term memory network to obtain the third time-delay compensation parameter; Step S50: Apply the third time-delay compensation parameter, input the third time-delay compensation parameter into the non-linear compensation filter to obtain the final time-delay compensation signal.

2. The data compensation method of a radio frequency chip according to claim 1, wherein In step S10, the step of real-time collecting the real-time status data of the I / Q signals of the RF chip, where the real-time status data includes real-time signal amplitude, real-time signal frequency, and real-time noise level, and calculating the first time-delay compensation parameter based on the fuzzy control algorithm combined with the real-time status data specifically includes: Step S101: Real-time collect the real-time status data of the I / Q signals of the RF chip at time t from the RF chip, including real-time signal amplitude A(t), real-time signal frequency f(t), and real-time noise level N(t); Step S102: Define three-layer membership degrees for the signal amplitude A(t), real-time signal frequency f(t), and real-time noise level N(t) respectively, and use the triangular method to quantify the signal amplitude A(t), real-time signal frequency f(t), and real-time noise level N(t) to form fuzzy variable inputs; Step S103: Deduce according to the preset fuzzy control rule base using the preset Mamdani model to obtain a fuzzy set output; Step S104: Perform defuzzification processing on the fuzzy set output, using the centroid method or the maximum membership degree method, and calculate the first time-delay compensation parameter.

3. The data compensation method of a radio frequency chip according to claim 1, wherein In step S20, the step of constructing a non-linear compensation filter, inputting the first time-delay compensation parameter into the non-linear compensation filter to obtain a preliminary time-delay compensation signal specifically includes: Step S201: Design a non-linear filter based on the Volterra series expansion method; Step S202: Input the first time-delay compensation parameter into the non-linear compensation filter, and perform weighted processing on the linear term and the high-order non-linear term in the non-linear compensation filter to obtain the mapped first time-delay compensation parameter; Step S203: Perform time-domain compensation on the mapped first time-delay compensation parameter to obtain a preliminary time-delay compensation signal.

4. The data compensation method for a radio frequency chip according to claim 3, characterized in that, In step S20, the step of performing time-domain compensation on the mapped first time-delay compensation parameter to obtain a preliminary time-delay compensation signal uses the formula: ； Among them, is the preliminary time delay compensation signal; is the first time delay compensation parameter after mapping; is the non - linear adjustment factor, which is used to further correct the compensation effect according to the signal estimation error; is the input signal; is the estimated value of the input signal; is the integral variable; is the error integral term; is the delay correction term.

5. A data compensation method for a radio frequency chip according to claim 1, characterized in that In step S30, the step of calculating the signal synchronization error according to the first time-delay compensation signal, and performing feedback adjustment on the first time-delay compensation parameter according to the signal synchronization error to obtain the second time-delay compensation parameter specifically includes: Step S301: Obtain the target synchronization signal from the preset prior information, and perform the same time benchmarking process on the target synchronization signal and the first delay compensation signal; Step S302: Calculate the signal synchronization error using the mean square error based on the benchmarked target synchronization signal and the first delay compensation signal; Step S303: Determine the direction and amplitude of the adjustment required for the delay compensation parameter based on the signal synchronization error in combination with the PID control method, and output a correction amount; Step S304: Apply the correction amount to perform feedback adjustment on the first delay compensation parameter to obtain the second delay compensation parameter.

6. The data compensation method of a radio frequency chip according to claim 1, characterized in that, In step S40, the steps of constructing a long short-term memory network and performing pre-training, and dynamically predicting the second delay compensation parameter through the pre-trained long short-term memory network to obtain the third delay compensation parameter specifically include: Step S401: Obtain the historical time series data of the second delay compensation parameter, the historical time series data of the signal amplitude, the historical time series data of the signal frequency, and the historical time series data of the noise level. Denoise, normalize, and time-align the collected time series data, and construct a fused feature vector; Step S402: Construct a long short-term memory network, including an input layer, multiple LSTM layers, a fully connected layer, and an output layer; among them, the input layer is used to receive the fused feature vector, and the multiple LSTM layers are used to capture the long-term dependencies and non-linear relationships in the time series data; the fully connected layer is used to map the output of the LSTM layer to the predicted value, and the output layer is used to output the predicted value; Step S403: Obtain the target synchronization signal from the preset prior information, construct a standard compensation parameter output based on the target synchronization signal, and train the long short-term memory network using the fused feature vector and the standard compensation parameter output; Step S404: Use the second delay compensation parameter as the output of the long short-term memory network, and the long short-term memory network outputs the third delay compensation parameter.

7. The data compensation method of a radio frequency chip as described in claim 6, characterized in that, In step S40, during the process of training the long short-term memory network, it also includes introducing a synchronization performance metric as a regularization term to design the loss function L, and the loss function L uses the formula: ； where \(i\) represents the \(i\)-th training sample, is the third time delay compensation parameter predicted by the long short-term memory network; is the standard compensation parameter output constructed according to the target synchronization signal; is a preset regularization coefficient used to balance the two parts of the loss; is a measure of the signal synchronization error after compensation based on the prediction parameter; \(N\) is the preset number of training samples.

8. A data compensation system for a radio frequency chip, applied to the data compensation method of a radio frequency chip according to any one of claims 1-7, characterized in that The data compensation system of the radio frequency chip includes: A real-time status data acquisition module, which is used to collect the real-time status data of the I / Q signal of the radio frequency chip in real time. The real-time status data includes the real-time signal amplitude, the real-time signal frequency, and the real-time noise level, and calculates the first delay compensation parameter based on the fuzzy control algorithm in combination with the real-time status data; A fuzzy control delay parameter calculation module, which is used to construct a non-linear compensation filter, input the first delay compensation parameter into the non-linear compensation filter, and obtain a preliminary delay compensation signal; A non-linear compensation filtering module, which is used to calculate the signal synchronization error based on the first delay compensation signal, and perform feedback adjustment on the first delay compensation parameter according to the signal synchronization error to obtain the second delay compensation parameter; A synchronization error feedback and parameter adjustment module, which is used to construct a long short-term memory network and perform pre-training, and dynamically predict the second delay compensation parameter through the pre-trained long short-term memory network to obtain the third delay compensation parameter; The LSTM time-delay prediction and optimization module is used to apply the third time-delay compensation parameter, input the third time-delay compensation parameter into the non-linear compensation filter, and obtain the final time-delay compensation signal.

9. A data compensation device for a radio frequency chip, characterized in that, The data compensation device of the radio frequency chip includes: a memory, a processor, and a data compensation program of the radio frequency chip stored on the memory and executable on the processor. When the data compensation program of the radio frequency chip is executed by the processor, it implements the data compensation method of a radio frequency chip according to any one of claims 1 to 7.

10. A computer program product, characterized in that, The computer program product includes a data compensation program of a radio frequency chip. When the data compensation program of the radio frequency chip is executed by a processor, it implements the data compensation method of a radio frequency chip according to any one of claims 1 to 7.

Images